We used to do this in 1.6.3 for a cluster of thin servers:

bundle exec thin --timeout 5 -C adapters/web_server/config/production_thin.yml restart

This was fine, because the --timeout wasn't passed into the server. It looks like that in 1.6.4, the servers started taking timeout configurations, so as of that version our command-line script wasn't just timing out the restart, it was setting a timeout config for the servers it restarted as well.

So we changed it to

bundle exec thin --wait 5 -C adapters/web_server/config/production_thin.yml restart

But this doesn't do what we want either; the --wait is filtered out of the start & stop jobs created by Thin::Controllers::Cluster using Thin::Command. I had started to work on a PR to pass --wait through from the cluster controller, but since it appears that it was omitted on purpose, I figured I'd file this issue to see if there was a better way, or if --wait was omitted because it was a problem.